Legal claims defining the scope of protection, as filed with the USPTO.
1. A method of image stabilization, the method comprising: determining whether an input image comprises a representative feature portion; sampling the representative feature portion to generate a sampled image if it is determined that the input image comprises the representative feature portion; enlarging the input image optically by a magnification factor and matching the enlarged input image with the sampled image or a resized sample image to obtain central coordinates of the enlarged input image, wherein the resized sample image is obtained by the magnification factor; and aligning an optical axis by calculating a difference between the central coordinates of the enlarged input image and the central coordinates of the sampled image.
2. The method of claim 1 , wherein the representative feature portion has a luminance level higher than a predetermined level or comprises a predetermined shape, and is selected from among a plurality of feature portions in the input image.
3. The method of claim 1 , wherein the determining whether the input image comprises a representative feature portion comprises: extracting a plurality of feature portions from the input image; selecting a candidate feature portion having a predetermined feature from among the extracted plurality of feature portions; and determining that the input image comprises the representative feature portion if the enlarged input image comprises the candidate feature portion, and determining that the input image does not comprise the representative feature portion if the enlarged input image does not comprise the candidate feature portion.
4. The method of claim 1 , wherein the matching the enlarged input image with the sample image comprises performing scale invariant feature transform (SIFT) matching if the magnification factor is equal to or less than 2.5.
5. The method of claim 1 , wherein the matching the enlarged input image with the resized sample image comprises performing template matching if the magnification factor is greater than 2.5.
6. An image stabilization system comprising: a zoom camera; an image preprocessor that converts an analog input signal input from the zoom camera into a digital signal; the system is further configured to: determine whether an input image comprises a representative feature portion; sample the representative feature portion to generate a sample image if it is determined that the input image comprises the representative feature portion; enlarge the input image optically by a magnification factor and match the enlarged input image with the sample image or a resized sample image to obtain central coordinates of the enlarged input image, where said resized sample image is obtained by resizing the sample image by said magnification factor; and align an optical axis by calculating a difference between the central coordinates of the enlarged input image and the central coordinates of the sample image.
7. The image stabilization system of claim 6 , wherein the representative feature portion has a luminance level higher than a predetermined level or comprises a predetermined shape, and is selected from among a plurality of feature portions in the input image.
8. The image stabilization system of claim 6 , further comprising: determining that the input image comprises the representative feature portion if the enlarged input image comprises a candidate feature portion, and determining that the input image does not comprise the representative feature portion if the enlarged input image does not comprise the candidate feature portion, wherein the candidate feature portion has a predetermined feature and is selected from a plurality of feature portions extracted from the image.
9. The image stabilization system of claim 6 , wherein the enlarged input image is matched with the sample image by scale invariant feature transform (SIFT) matching if the magnification factor is equal to or less than 2.5.
10. The image stabilization system of claim 6 , wherein the enlarged input image is matched with the resized sample image by template matching if the magnification factor is greater than 2.5.
Unknown
May 12, 2015
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.